HIGH SPEED MAGNETIC FIELD

GENERATOR
Design Document

Advisers:
Dr. Mani Mina
Wei Shen Theh

Team May1740:
Patrice Bruce
Paul Danker
Michel Lemoupa Fokou
Dodji Monglo
Jacob Sinn

Revision 2.0
Last Edited:10.20.20
TABLE OF CONTENTS
1 Introduction..............................................................................................................................4

Problem Statement.........................................................................................................................................................4

Project Purpose...............................................................................................................................................................4

Project Goals...................................................................................................................................................................4

2 Deliverables..............................................................................................................................4

Fall 2020 – EE491............................................................................................................................................................4

Spring 2021 – EE492........................................................................................................................................................5

3 Design.......................................................................................................................................5

System Specifications......................................................................................................................................................5

Non-functional...........................................................................................................................................................5

Functional...................................................................................................................................................................5

Proposed Design.............................................................................................................................................................5

Resistors.....................................................................................................................................................................6

Capacitors...................................................................................................................................................................6

Diode..........................................................................................................................................................................6

Inductor......................................................................................................................................................................6

MOSFET......................................................................................................................................................................7

Design Analysis................................................................................................................................................................7

Current through the inductor (transmitting coil).......................................................................................................7

Simulation software limitations.................................................................................................................................8

4 Testing and Development.......................................................................................................10

Software and Hardware................................................................................................................................................10

Software...................................................................................................................................................................10

Hardware.................................................................................................................................................................11

5 Measurements.........................................................................................................................11

2
 Coil 1 Test.................................................................................................................................................................12

Coil 2 Test.................................................................................................................................................................12

6 Future Work...........................................................................................................................14

Low-Power Design....................................................................................................................................................14

Current-Sense Resistor.............................................................................................................................................15

7 Conclusions............................................................................................................................16

8 References..............................................................................................................................16

9 Appendix................................................................................................................................17

Importing external SPICE file to NI Multisim............................................................................................................17

1INTRODUCTION

PROBLEM STATEMENT

Magnetic field pulse generators have been used for various applications across many different electrical
engineering disciplines. Technical requirements of generator systems vary based on the specific application. One of
these applications is magneto-optical (MO) switching within fiber optic networks. These networks currently use
optical switches and routers to direct signals to desired paths defined by electrical energy. During this process, the
conversion from optical energy to electrical energy causes problems. Due to the different bandwidths between
optical systems and electrical systems, the signal is not directed to its desired path as quick as needed. However,
by utilizing the magneto-optic effect (Faraday effect), the router switching can be achieved without optical-
electrical conversion and, consequently, without the high switching latency. One way of inducing the MO effect on
the switching system is by applying a strong and quick magnetic field pulse. Past circuit designs have shown that it
is possible to create pulse generators with relatively high power and speed. However, there is room for
improvement in pulse strength, rise time, and power efficiency.

PROJECT PURPOSE

The purpose of our project is to design and implement a circuit that generates a quicker, stronger pulse while also
improving power efficiency. This would provide a more practical way to use magnetic pulse generators in optical
switching and higher-scale applications such as transcranial magnetic stimulation (TMS).

PROJECT GOALS

The goals of this project are fourfold:

1. Gain a deep, applicable, and robust understanding of magnetic pulse generator concepts and circuit
theory that will help us in our future careers and education.
2. Build a circuit that can generate a 500 Gauss magnetic field pulse with a rise and fall time of at most, 10
nanoseconds.
3. Design and build a low-power version of the pulse generator
4. Identify a more effective method of directly measuring the current through the circuit coil

2DELIVERABLES

The deliverables for our project will be divided between first semester (EE491) and second semester (EE492).
Completing these according to our semester timeline will keep us on track to achieve our goals by the end of the
year (Spring 2021).

FALL 2020 – EE491

Our final review presentation will cover learning circuit theory and fundamentals, developing PCB fabrication skills,
and new circuit modifications/improvements. We will also have two functional circuit boards with us at the final
review:
 1. A board identical to SdMay21-42 group circuit
2. A board with testing points for different current sense resistors

SPRING 2021 – EE492

In April, we will give a presentation on our progress throughout the Spring semester. This will primarily consist of
how we built upon the foundation that was established in EE491. We plan on outlining the implementation of our
circuit modifications and presenting our solutions to specific problems such as power consumption and
measurement techniques. We hope to demo our low-power solution and show how our circuit modifications make
test and measurement more accurate and consistent.

3DESIGN

SYSTEM SPECIFICATIONS

NON-FUNCTIONAL

1. The product must be on a PCB with a footprint of less than 3.5” x 2”

2. The product output must consistent
3. The product must be fitted in an insulating enclosure
4. The product must have quality soldering for long time using

FUNCTIONAL

1. The product shall output a magnetic field pulse peaking at 500 Gauss with rise time of less than 0.15 𝜇s
2. The DC power supply must not exceed 15 V
3. The transmitting coil must be fitted for the fiber optic cable

PROPOSED DESIGN

Figure 1
RESISTORS

1. The 2 ohm resistor (Figure 1 - R3) placed series with the diode is used to dissipate some of the energy in
its loop. Because there is such a large amount of current flowing through the inductor, this resistor is
needed to stabilize the circuit.
2. The resistor in parallel with the pulse source (Figure 1 – R2) is used to match the output impedance of
the source. For this source, we are using a Tektronix AFG 3021B function generator which has a 50 ohm
output impedance. Matching this impedance is vital in order to prevent signal reflection. In other words,
this causes all of the power output of the source to reach the gate of the MOSFET (ideal case).
3. The resistor connected to the MOSFET source (Figure 1 – R1) is used to control the current through the
inductor. Because the inductor path can be characterized as an LR circuit, its current pulse characteristics
can be controlled through R1. This resistor is also used to measure the current through the inductor (in
series with the inductor when the MOSFET is “on” (see Future Work section).

CAPACITORS

In order to produce such a high current through the inductor, the circuit needs a dedicated capacitor bank. Our
circuit contains a bank of four capacitors in parallel, creating 200.2 𝜇𝐹 of energy storage. The two larger capacitors
(C1 and C2) are primarily used to store the charge from the DC source that is eventually sent through inductor,
creating the magnetic field. The two smaller capacitors (C3 and C4) act as protection against harmful current
spikes. Because the DC source is only 15 V, this capacitor bank is essential in creating the necessary current flow
through the coil.

DIODE

The diode placed in series with the resistor can seem as a voltage regulator. We use this diode to make our output
stable. The diode can keep a constant voltage at the inductor, so the current through the inductor can be stable, so
that the magnetic field pulse from the inductor will be better.

INDUCTOR

The inductor is one of the most important components in our circuit. When calculating the coil’s specifications, we
have referenced the bottom two equations. Our project requirements are to generate a magnetic field (B) pulse
peaking at no less than 500 Gauss with a rise time of no more than 0.15 𝜇𝑠. Combining these two requirements
essentially defines our current and inductance levels for us, therefore leaving us with specific coil dimensions. For
further analysis on this concept, please refer to the Future Work section of this document.

𝜇𝑁𝐼 𝜇𝑁2𝜋𝑅2
𝐵= √𝑙 +4𝑅2
2 L = √𝑙2+4𝑅2
B = magnetic field (Teslas) 1Tesla = 10000 Gauss I = current through the coil (Amperes)
L = inductance (Henries) 𝑙 = length of the coil (meters)
µ = permeability of free space (4𝜋 × 10−7 𝐻/𝑚) R = radius of the coil (meters
N = number of turns in coil
MOSFET

Throughout this semester we have used the same MOSFET that the May1740 group used over the past year. This
led to an easier learning curve because we could get personal help from them. There are certain characteristics of
the current MOSFET that are undesirable in some of our future goals. Because of this, we are going to be exploring
other MOSFET options in the future, specifically with the low-power solution (see Future Work section).

DESIGN ANALYSIS

After researching the concepts behind our circuit, we began simulating the May1740 group final circuit. To do this,
we had to import the PSMN1R2-30YLD MOSFET made by NXP Semiconductors. This turned out to be quite a
challenge because we are using NI Multisim 14.0 as our circuit simulation engine. Mulitsim does not provide a way
to direct import components using a SPICE file (see Appendix). Once we placed the FET, building the circuit was
easy and we were able to move onto “playing to learn”. In order to gain a deeper understanding of the circuit’s
functionality, it was vital that we messed with components, changed simulation profiles, and used a variety of
sources. This led us to really understand aspects of our circuit like: why we need a diode, how does the MOSFET
work and contribute to our goal, and how our inductance and capacitance values dictate the magnetic field output
magnitude. Throughout this experimental time, we recorded many scope images that show the behavior of the
circuit configuration. In this section, we will outline a few of our most significant findings.

CURRENT THROUGH THE INDUCTOR (TRANSMITTING COIL)

Figure 2
In this graph (Figure 2), you can clearly see the difference between the current drawn through the two different
inductors (146nH and 146pH). Although both achieve the desired current level of 50 A, the 146pH inductor has a
significantly smaller rise time which is vital for our application. The higher latency of the higher inductance would
cause slower switching in an optical fiber network, which is what we are trying to solve. The reasons for these
results are somewhat intuitive. Because of the higher inductance, the 146nH indicator takes a lot longer to charge
and discharge the current. This change can easily be calculated by using the LR circuit time constant equation:
L
𝜏 =
𝑅
This equation shows the direct relationship between inductance and transient time. It is also known that it takes
about 5T to reach steady-state behavior. The challenge of our project is going to be finding a way to achieve a fast
rise time while maintaining a sufficient current through the inductor to generate the strong magnetic field. Ideally,
when only looking at rise time, smaller inductance is better. However, this decreases the ability to output strong a
magnetic field. Furthermore, the physical limitations of our application (fiber optic switching) require a minimum
coil diameter which makes 𝑝𝐻-scale inductances virtually impossible.

SIMULATION SOFTWARE LIMITATIONS

As we experimented with different circuit configurations we found that Multisim has various limitations that make
observations difficult to rely on in certain instances.

Our first experience with these limitations came when testing the effect of removing the shunt resistor at the
source of the FET. Removing this resistor yielded a pretty ridiculous graph (Figure 3) that shows the inductor
current peaking at over 1200 A. We know this is impossible for our circuit in real application so we think Multisim
is showing its limitation in dealing with truly ideal situations (ideal short between FET source and ground). In order
to solve this problem, we attempted to limit the current from the DC course by placing a 0.75 ohm resistor in
series (Figure 5). This did, in fact, limit the current of the circuit without the FET source resistor (Figure 4).
However, the results show a highly unstable behavior with slow rise and fall times.
Figure 3

Figure 4
 Figure 5

Lastly, we have observed a limitation in showing the effects of the charging and discharging of our capacitor bank.
We know that our capacitors are meant to store and release charge from the DC source. However, removing the
entire capacitor bank from our circuit yields the same exact plot for the current through the inductor as with the
capacitors in place. This shows Multisim’s inability to show the real-time active behaviors of certain elements.

As we move forward with simulations, our goals are going to be adjusted in order to explore different avenues
than the Dec1622 group. One of our main ideas is developing a low-power version of the magnetic field generator.
We have a few preliminary ideas on how to start on that new path. The big idea is to try to modify the coil in such
a way that allows us to draw less current and potentially use lower-powered sources.

4TESTING AND DEVELOPMENT

SOFTWARE AND HARDWARE

SOFTWARE

Past teams working on this project have used a variety of different software platforms to test and simulate circuit
configurations. For our project, we decided to use National Instruments Multisim 14.0 for circuit simulation and
National Instruments Ultiboard 14.0 for PCB layout. Our team, in general, has more experience using Multisim
from past coursework as opposed to OrCAD PSPICE (used by the Dec1622). Another reason why using PSPICE
would not work is that most of our teammates PCs run Windows 10 which causes PSPICE to crash. Our advising
team recommended using a Windows 7 PC if we chose to use PSPICE, but eventually decided to go with Multisim.

Once we reached the decision to use Multisim as our simulation platform, we immediately ran into some setbacks.
First, we found that there is no direct way to import SPICE file components into a Multisim model. To recreate the
Dec1622 group’s circuit, we needed to be able to import a MOSFET from NXP Semiconductors which we found on
their website in a .lib file format. However, we soon found that it is not possible to load the library file and drop
the new component directly into our model. Through extensive online research, we eventually found a
roundabout way to use external SPICE components in Multisim (see Appendix).
After we decided on the circuit simulation software, it was easy to choose a layout design software. National
Instruments has a layout software that works seamlessly with Multisim circuit design called Ultiboard.
Furthermore, some of our team members have experience using Ultiboard from EE 333 coursework.

Lastly, we will use MATLAB to build a low-power design model. Ideally, we will be able to graphically show that
there is an optimized level of current for a certain time constant. This will show that there is a way to use lower
current, and therefore lower power, while maintaining an acceptable rise time.

HARDWARE

We will use all of the standard EE lab equipment to test and evaluate our circuit. Our equipment includes the
following:

 Tektronix AFG 3021B Single-channel function generator

 Agilent E3631A DC Power Supply
 Tektronix DPO 4032 Digital Phosphor Oscilloscope
 Various 50 ohm impedance coaxial cables
 Handheld Multimeter
 ISU ECpE Fabrication Drill

5MEASUREMENTS

Once we had our circuit fabricated, we spent some time testing and measuring its actual behavior. Using
simulation software is just a foundation for design, but cannot be relied on when it comes to real-world behavior.
Through testing and measurement, we were able to get a clear look at the non-ideal behavior of our circuit. Figure
6 shows our experimental setup for these tests.

Figure 6 – test setup

COIL 1 TEST

For our first test, we designed a coil with 5 turns and and inductance of about 139 nH (figure 7). This test yielded
the pulse seen in figure 8. As a team, we were excited to see a clean plot that shows relatively predictable
behavior, however we were not thrilled with the slow rise time.

Figure 7 – coil 1 parameters

Figure 8 – coil 1 pulse profile

COIL 2 TEST

After seeing decent results for our first coil, we wanted to design a second, smaller, coil (figure 9). We knew that a
lower inductance would yield a shorter, more desirable, rise time (figure 10).
 Figure 9 – coil 2 parameters

Figure 10 – coil 2 pulse profile

 Figure 11 – Main Circuit Board Figure 12 – Main Circuit Board Layout

6FUTURE WORK

Now that we have the foundational concepts of our project, we are confident that we can make significant
contributions to the improvement of the magnetic field generator for future MO switching applications. As our
project has developed, we have constructed ideas for future improvements to be worked on next semester.

LOW-POWER DESIGN

Fiber optic networks are characterized by rapid data rates with low latency at high efficiency. In order to improve
the overall performance of the system, it is necessary for our circuit (MO switch) to be operating at maximum
power efficiency. In our application, there are various ways to change the power drawn by the circuit. First, and
perhaps the most obvious, is to modify the power sources used in our model. Decreasing the source voltage clearly
decreases overall power consumption. Secondly, we can modify other circuit components (resistors, capacitors,
MOSFET) in order to draw less current through the coil. This is the method that we are going to pursue next
semester because it gives us the most freedom for modifications.

We have started initial calculations to determine the feasibility of this solution. Through this analysis, we believe
that we will be able to adjust circuit parameters in order to draw less current through the coil while maintaining
pulse magnitude and rise time. Please view our calculations below.

L
𝜇𝑁𝐴 𝜇𝑁2𝐴
Base Equations: 𝜏= 𝐵= L =
𝑅 √𝑙2+4𝑅2 √𝑙2+4𝑅2
2
𝜇𝑁 𝐴
𝜇𝑁 L L𝐼 𝐵𝑁𝐴 𝑩𝑵𝑨
Calculations: L=  = 𝐵= L= 𝝉=
√𝑙2+4𝑅2 √𝑙2+4𝑅2 𝑁𝐴 𝑁𝐴 𝐼 𝑰𝑹

Conclusion: 1. 𝑐𝑜𝑛𝑠𝑡 𝜏 = 500(𝑁𝐴)↓ 500𝑁𝐴 500𝑁𝐴↓

𝐼↓ 𝑅
2. 𝑐𝑜𝑛𝑠𝑡 𝜏 = 𝐼↓ 𝑅↑
3. 𝑐𝑜𝑛𝑠𝑡 𝜏 = 𝐼↓ 𝑅↑
Example: say we want to have a 100 nanosecond rise time. This would make 𝜏 = 20 𝑛𝑠 since 𝑟𝑖𝑠𝑒 𝑡𝑖𝑚𝑒 ≈ 5𝜏. To
achieve this goal while lower the current through the inductor, we can decrease the dimensions of the coil
(method 1). Another way of accomplishing this is to increase the resistance of the circuit through increasing the
MOSFET channel resistance and/or the source resistance (method 2). Furthermore, we can combine these two
methods to more aggressively counteract the lesser current (method 3).

There is no doubt that we still have a lot of work to do in this area next semester, but we believe this solution
shows promise and opens the door to other key circuit modifications with the MOSFET and source resistor.

MOSFET: We are finding that the current MOSFET may present challenges to the low-power solution. Tests of our
circuit have shown that the FET is operating in its triode region for about 33% of the time and saturation region
66% of the time. Ideally, we could push the FET to saturation in order to tune the channel resistance to a suitable
(higher) value for our application. However, we are seeing that in both saturation and triode regions the channel
resistance is on the scale of milli-ohms. Because of this, it seems impossible to tune the FET for our low-power
scenario. Next semester, we will move forward by searching for a suitable MOSFET for this solution.

Source Resistor: Our calculations show that we can achieve lower-power solution by increasing the overall
resistance of the circuit. As we showed in the previous section, this can be achieved by modifying the MOSFET
channel resistance. However, it can more directly be accomplished by increasing the value of the resistor
connecting the FET source to ground. Right now, our circuit is generating about 80 W of power in the source
resistor that is only rated for 25 W. This poses a problem that we think is causing false data in our measurements.
Next semester, we will find a larger resistance rated for a higher power.

CURRENT-SENSE RESISTOR

The current circuit uses the MOSFET source resistor as a current-sense resistor. However, we believe that placing a
small resistor in series directly next to the inductor would provide a more accurate way of measuring inductor
current and, therefore, magnetic field strength. We are working on a test board layout that leaves two openings
for resistors: one next to the inductor and one connecting the FET source to ground. This will give us an easy way
to test our hypothesis that the new resistor placement will lead to better test results. Please find the circuit board
and layout of this design below.

Figure 13 – Current Sense Board Figure 14 – Current Sense Board Layout

 7CONCLUSIONS

In conclusion, we have now caught up to last year’s group on our circuit’s concepts, functionality, challenges, and
purpose. We thoroughly understand the fundamentals of how the circuit works and why we use the components
that we do. Now, we are prepared to move forward into the stage of identifying improvements and implementing
them. We are confident that by the end of Spring 2017 we will have a new and improved functioning PCB of this
magnetic field generator that has been worked on for years.

8REFERENCES

1. NXP Semiconductors. (2015). PSMN0R9-30YLD Product Data Sheet [PDF]. Retrieved from
http://www.nxp.com/documents/data_sheet/PSMN0R9-30YLD.pdf

2. "EE 333 : Lab." EE 333 : Lab. N.p., n.d. Web. 11 Mar. 2016.

http://tuttle.merc.iastate.edu/ee333/info.htm

3. Pritchard, John W., Mani Mina, and Robert J. Weber. "Magnetic Field Generator Design for Magneto-Optic
Switching Applications." IEEE Trans. Magn. IEEE Transactions on Magnetics 49.7 (2013): 4242-244. Web.
“””
4. Simple FET DC Bias Circuit (2016) [PDF] Retrieved from

http://my.ece.ucsb.edu/York/Bobsclass/2B/Extras/FET%20Biasing.pdf

5. "How Can I Import a SPICE Model Into Multisim?" - National Instruments. N.p., n.d. Web. 06 Nov. 2016.

http://digital.ni.com/public.nsf/allkb/2373E75D8B375EA1862575D2004D9C88

6. Razavi, Behzad. Design of Analog CMIS. Los Angeles: U of California, 2001. Https://www.u-
cursos.cl/usuario/9553d43f5ccbf1cca06cc02562b4005e/mi_blog/r/[Razavi]_Design_Of_Analog_Cmos_Int
egrated_Circuits.pdf. Web. July 2000.
7. Pritchard, John W., Mani Mina, Robert Y. Bouda. “Feel the pulse”, Magnetics Technology International,
p4-7 (2013).

http://viewer.zmags.com/publication/17fde0ad%23/17fde0ad/6

8. "Magnetic Pulse Generator." N.p., n.d. Web.

http://dec1622.sd.ece.iastate.edu/

9. Pritchard, John W., Mani Mina, and Narimdinada Y. Bouda. "Feel the Pulse." N.p., n.d. Web.

http://viewer.zmags.com/publication/17fde0ad#/17fde0ad/6
9APPENDIX

IMPORTING EXTERNAL SPICE FILE TO NI MULTISIM

1. Search your package number of MOSFET on google.

2. Download the SPICE model on the website
3. You will get a page of data about this MOSFET, copy all of them into a new text on your desktop.
4. Open NI Multisim
5. Select Place->Component
6. Copy and paste the package number to the highlighted place

7. Delete the last “D” and find a similar part

8. Click the OK button to place it on the work area

9. Double left click on the component on the work area and select the Edit Component in dB button

10. In the component properties window, select the Model tab and then press Add/Edit button
11. Under Database Name select User Database and then press the Add button

12. Type your package number in the Model ID field and click OK
13. Copy the data you saved in the text and paste it to the window under Model data

14. Click on the Select button and click the Yes button to change the model data

15. Under Model Name select the original model and click the Delete a Model button
16. Click on the General tab, type your package number in the Name field

17. In the select Destination Family Name window, select User Database and choose Transistors, and then
click Add family, next click OK
18. Now when user Place->component, in the user database, the component will be showed and user
can place it to the work area